So this clever new idea has been floating around in my head for a couple of days now. If you consider a router, it’s basically connecting directing packets to different subnets. The simple case is when you have two subnets and a router that’s in the middle.

Let’s call the router R1, with the two subnets being S1 and S2. Let’s say you’re managing a bunch of servers in a data center rack and they’re all on S2. Your router, R1, connects them to your bandwidth provider who is on S1.

R1 has two interfaces – one for S1 and one for S2 – and it has an IP address on each. Your provider is routing all of the traffic going to S2 to R1’s address on S1. All of your hosts on S2 are sending the packets that need to go outside the subnet to R1’s address on S2.

This is really simple (conceptually, of course). I think you can write something that can do this with a couple of raw sockets. Well, I think I can actually write something that can do simple routing like this. All you have to do is peek at the IP headers and rewrap the IP packets with new Ethernet headers.

Argh, I don’t think I have the hardware to test this out easily. Maybe I can play around with some VMs? I haven’t tried that out yet.

For some reason I keep thinking that the kernel doesn’t like this. Perhaps iptables will mess with stuff or something. Anyway, it’d be really cool to write a firewall in Go that does some neat deep packet inspection.

Increasing my knowledge of the networking stack :D